Positive users applaud Substack's Pangram AI detector for curbing unbearable spam and protecting human writing, while negative users call it biased, inaccurate, or hypocritical.
Based on 149 visible X reactions from 272 accounts.
Ask a question below.
Published answers will appear here.
The click-to-reveal tool estimates the proportion of AI-generated text.
What an idiotic and backward move. I want to read texts that are GOOD, whether they are written partly by AI or not. The "AI= always slop" equation is a stupid 2023-ish take. What is REALLY badly needed is a general anti-BS, anti-Slop filter that filters BAD writing, regardless if the author is human or AI.
@Substack @pangram PANGRAM IS A PANSCAM! i was able to fool it in just two attempts by asking sonnet 5 to "write like a human" what is the point? there are no patterns to AI writing for you to detect this fundamentally cannot work stop scamming people, NOW! https://x.com/RealAstropulse/status/2070481628325876027
@max_spero_ Insanely cool stuff today @max_spero_ - congrats. And keep it up.
Bravo. Chris & team have really created a refuge for people who think in public, warding off threats to independent writing from censorship to, now, the slopocalypse “The problem is not people using AI, it’s Claudefishing; the deception when you think you’re getting a human connection but there’s no one on the other end. The problem for the platform is that when “content” made by no one takes over, it drives out human voices and pollutes the commons. When readers have to wonder if what they’re reading is real, it undermines trust in authorship and threatens the livelihood of writers—including those who use AI tools thoughtfully to produce work they believe in. Platforms that reward fakeness will drive a race to the bottom. @pangram’s data shows that as much 40% of long-form text on LinkedIn is generated. We don’t have all the answers, but we don't want to wait to act until it's too late. So as a first step we’re adding the ability to scan to give some transparency and a basis for feedback: what do you want to see?”
Positive users applaud Substack's Pangram AI detector for curbing unbearable spam and protecting human writing, while negative users call it biased, inaccurate, or hypocritical.
Based on 149 visible X reactions from 272 accounts.
Ask a question below.
Published answers will appear here.
Substack will now let you see to what degree something is written by AI, using Pangram. Readers should not be burdened with the effort of reading if writers don’t put in the effort to write: it’s a breach of the social contract of the internet. I use the Pangram chrome extension for this for LinkedIn / X, but it’s awesome to see big platforms adopting it as a first party integration!
substack's adding pangram integration, and i really like both the initial intervention - click-to-reveal - and the described (uncertain!) path to future interventions, like filtering out anything that scores highly. seems thoughtful & measured. https://substack.com/home/post/p-207812763
This is an interesting experiment (and I am glad that I have always written my own posts!)
it'd be absolutely hilarious if linkedin actually implemented this.
not to flex, buuuut